Mainly to the ZZ variant, ZZ D its on the speedsolving wiki you should read about it, it's interesting.
I have and I can pull it off in linear FMC, but I don't have a clue about how to recognize the last 6 corners for a speedsolve. That is where ZZ-rainbow shines. You still have 2gll, but you also have an ok recognition. I knew my 2gll, but I am starting to forget it. It was so easy to learn with all of the sune combos, But I am starting to run into cases that go something like this... "

I know it is an antisune, but is it on the right or left? :fp". This is a big problem I have with some Algs. When I stop using them and practicing them they start sliping from my mind.
if you're that interested in it, i'm hoping to see your developments in it since ZZ-d is pretty unexplored since noone has been able to find a decent way to detect corner permutations while solving first block, do that and we're golden. if you're really that interested in it, also check out ZZ-rainbow, it's somewhere in the ZZ/ZB thread. everyone will legitimately hug you if you missing link zz-d brobro
This.
What variants are there? I know of ZZ-a ZZ-b ZZ-d. What else and what are they.
Well you have the obvious OLL/PLL, COLL/EPLL (zz-vh) but there are many other methods I have read about when EO is good and you reach the last slot.
ZZ-rainbow solve CP in 5 moves before the last slot/ 2GLL
ZZ-C is winter variation, but you get an ocll skip from any f2l case
MGLS-Z (only the CLS step) part of ZZ-C
ZZ-WV winter variation Part of ZZ-C
The pairs approach of the heise method. build 2 pairs,solve last 5 edges and 2 corners
ZZ-Blah you disorient the corners and solve with the ZB Pi and H cases
OCELL/CPLL orient corners and solve edges/ permute corners
Snyder method's reduction to L4C
Build a block on top and use tripod cases
That and what you already said is what comes to mind right now, but it is likely that I missed something. At the end of the day, the easy stuff like Coll/Pll/WV stay at the top because they are proven to work very well.